All systems operational

Processing queues have drained and latency is back to baseline. A downstream model-provider slowdown was the root cause; we have added surge capacity to the pipeline.

Additional workers deployed. Queue depth is falling and we are monitoring recovery.

Root cause identified as elevated latency from an upstream AI provider. Document uploads succeed but classification and extraction are delayed.

We are investigating slower-than-normal document classification. No data loss; queued documents will process once throughput recovers.

Maintenance completed successfully. All systems are fully operational.

Planned Postgres version upgrade begins. The dashboard may be briefly read-only during failover. Filings and processing resume automatically afterward.

A bad deploy was rolled back and error rates returned to normal. We have added a canary check to catch this class of regression before full rollout.

A recent deploy introduced elevated 5xx responses on a subset of dashboard actions. Rolling back now.

We are investigating a spike in errors affecting some dashboard actions.